Dartmouth College recently installed a $2.5 million brain-analyzing machine because they want to answer important questions, like whether it’s scarier when the giant boulder chases Indiana Jones or when the Nazis’ heads melt. And they’re not going to let the dead-fish debate get in the way.
Confused? Understandable, since that may be the most baffling introductory paragraph I’ve ever written (No, it’s not – editor).
So let’s parse things a bit. First, we’ll deal with Indiana Jones, then we’ll get to the dead fish.
“Watch Raiders of the Lost Ark, we can tell what part of that movie the person is watching because of the pattern of their brain, as compared to during the rest of the movie,” said James Haxby, director of Dartmouth’s Brain Imaging Center, which uses a functional MRI scanner to peer inside subjects’ gray matter as part of the Center for Cognitive Neuroscience.
The center recently upgraded its functional Magnetic Resonating Image scanner in the basement of Moore Hall to a more powerful version rated at 3 Tesla, which is a measure of magnetic field strength rather than of automobiles. They opened up two walls and used a crane to install the 26,000-pound device, not just to do more neuron-level film reviews but also to continue Dartmouth’s long history of research via MRI.
Magnetic Resonating Image scanners use magnets to align hydrogen atoms in water molecules throughout our body, then take the resulting release of energy to create images of soft tissue and blood flow. Functional MRIs are specialized variants that focus on blood flow in the brain.
In 1999, Dartmouth became the first college to install an MRI scanner in a non-medical setting, strictly for research. These days, the functional MRI is used by some 15 principal investigators, with grad students, post-docs and undergraduate students on their team. That’s about 50 to 60 people in all, Haxby said.
Researchers want MRIs to help answer questions about how the brain works physically – how it’s wired – but also how it works psychologically – how it makes us think.
“We’re trying to understand the brain’s basis of personality differences, trying to understand how the brain understands other people,” said Bill Kelley, a Dartmouth professor of psychology who will take over as the new director of the Brain Imaging Center this year as Haxby steps aside.
The old scanner did this, too, but the new one will do it with more precision.
“We have a pretty good map of the parts of the brain that are involved in brain perception, now we’ll be able to look into these brain areas, decode it better,” Haxby said. “It’s like trying to see the landscape on one of the moons of Jupiter. With a better telescope you’re going to be able to see the details.”
Okay, that’s clear enough. But what was all that talk about dead fish?
That phrase is shorthand for an ongoing discussion about the effectiveness of MRI-based brain analysis. It comes from a 2009 study by a then-Dartmouth grad student, Craig Bennett, and a few others.
They took a dead Atlantic salmon and placed it in Dartmouth’s fMRI machine. At this point, I cannot improve on the description in their research poster: “The salmon was shown a series of photographs depicting human individuals in social situations with a specified emotional valence. The salmon was asked to determine what emotion the individual in the photo must have been experiencing.”
Their analysis of the resulting MRI images found that enough voxels – the MRI equivalent of pixels on a computer screen – lit up to indicate that the dead fish was reacting to some pictures.
Except that it wasn’t, of course, as the researchers said. Their work (which, you won’t be surprised to hear, won a science-humor Ig Nobel Prize) had a different goal, Kelley said.
“The dead fish study basically showed that if you do your statistics incorrectly, you can show (brain) activation in a dead fish. They purposely did not do the statistics correctly,” Kelley said.
Or, more succinctly, they showed this: “If you do bad science, you get crummy results.”
Statistical analysis is needed because an MRI image isn’t like a Polaroid picture. It’s the visual representation of huge amounts of data – 30,000 to 50,000 voxels, each altering or not altering over time, depending upon magnetic fields.
“With that much (data) . . . it’s very easy to have some fluctuation somewhere that’s just chance. Proper statistics determines that you don’t interpret a chance finding as a real finding,” Haxby said.
The issue is what constitutes proper statistics.
A recent study from researchers in Sweden and the U.K. claimed that improper statistical analysis occurred in the software used in up to 40 percent of past MRI studies, meaning many may have reported brain activity where none actually occurred.
Haxby and Kelley say the concern is overblown, although it continues to be much debated in the field.
They agree, however, that there’s one issue. Because it costs $500 to $1,000 of scanner time for each run, budget constraints limit study size, sometimes making it hard to analyze what happens or to reproduce results.
“One criticism of brain imaging is fair: We often don’t have as much data as we’d like. You’d like to study 1,000 people and can only study 50,” Kelley said.
The online world and open access may change that, he said. “The field is starting to move toward larger data sets, due to data sharing and data mining.”
If so, behemoths like the new scanner in Hanover may lead to more insight about what makes us tick. Maybe they’ll even be able to explain why Indiana Jones and the Crystal Skull was so boring.
(David Brooks can be reached at 369-3313 or dbrooks@cmonitor.com or on Twitter @GraniteGeek.)